New Cameras - The Evolution of HD

[mattharding]

I am a bit new to the world of HD. I have shot only a dozen or so projects on the F900 and Varicam. The new development of cameras, such as the Dalsa Camera on the homepage prompt me to ask some questions:

 

1) First, it seems that the cameras like the Viper and Dalsa are going in a completely different direction from the Sony/Panasonic world with the optical viewfinder, larger CCD for 35mm depth of field, etc. Why are Sony and Panasonic not quicker to embrace these film-style characteristics?

 

2) Why did Panavision settle on a sort of "middle of the road" with the Genesis? There is one large 35mm size CCD but there is no optical viewfinder and the recording device on the camera cannot record nearly all the resolution the CCD is putting out?

 

3) And finally, what is the mainstream high-end motion picture HD camera system going to most look like in the future? Not too long ago Lucas and his camp were talking like they had arrived with the F900 and now its specs look almost laughable compared to the new stuff out.

 

I have a personal theory that the HD world might split into two eventually. The ENG/HD world with an electronic viewfinder, 2/3" chips (because a shallow DOF and bigger lenses would be bothersome when shooting a home-improvement show, for example) and a recording devise onboard where compression is acceptable and welcome because it brings with it ease of use and quick work-flow. And then the high-end cinema HD with optical viewfinder, large CCD, no recording devise on board for the massive amount of data required to store an image that approaches film resolution, and a more complex work flow that will make it harder to deal with the huge high-res data files. Sound right?

 

I do believe that some day, this high end digital technology will win over the hearts of even the most discerning eye. That seems to be quite a way in the future. But what is so funny is watching companies like Dalsa and Viper BENDING OVER BACKWARDS, going to through the greatest lengths to try to be competitive with the current film technology. And after tons and tons of R & D that Dalsa went through, the best they can try to shoot for at their side by side comparison in Canada is to be "as good as" the newest Vision stocks out there. It is just funny to see people so passionately pursuing a new technology that can only hope to produce results as good as the current technology leader. Think of how far we are from reaping the benefits of using the Dalsa, fore example, over film. New untested work flowlow, different post production, unproven and uncertain results.....I guess you have to start somewhere......

[David Mullen ASC]

There is a belief by some -- mostly film people of course -- that digital moviemaking has to break free from its ENG origins. Cameras like the Genesis, Kinetta, Dalsa, Arri D20 are steps in that direction. The Viper however is still designed like a traditional video camera (but not a camcorder.)

 

Panavision felt that it was important to simplify the recording process and keep the Genesis free from being tethered to a hard disk recording system, the only other way of capturing 4:4:4 HD, so they based the recording end on the portable HDCAM-SR deck, which can record 4:4:4 HD with mild compression. The specs say that it can support up to 60P at 1920 x 1080.

 

The camera has a 12MP single CCD so it probably can send out even more resolution and data than the HDCAM-SR recorder can handle, but then you're stuck with finding a way of recording that much data in real time. For example, the Dalsa captures a 8MP image but uses recorders the size of mini-refrigerators that barely hold more than a half-hour of footage at that data rate. So sticking with the new HDCAM-SR tape format and building the deck into the camera was a decision by Panavision to make the camera more convenient to use, even though this limits the image to 1920 x 1080 pixels. However, that may change in future cameras.

 

Despite all the breakthroughs in non-tape recording, like hard disk recorders, tape still manages to be the most convenient and cost-effective way of recording and storing many hours worth of footage.

 

Obviously there will always be a need for ENG-style shooting in HD, so it is not unreasonable to see a split in types of shooting technology, one that resembles film cameras more while others resembling ENG cameras.

 

ENG video camera sales FAR outnumber 35mm film camera sales; Sony and Panasonic are based on mass production of components and cameras to make a profit, unlike a film camera company like Arri or Panavision, which more or less hand-build a small number of cameras each year. They don't stamp them out quickly on an assembly line in the same way a video camera is made. It's hard for Sony or Panasonic to get beyond the 2/3" CCD approach for pro video cameras since they would not be mass producing 35mm-sized imagers to the same degree (also ENG shooters might not want that large of an imager if it made the camera bigger or more expensive.)

[Phil Rhodes]

Hi,

 

For the record, HDCAM-SR uses the "studio profile" I-frame MPEG-4 codec. 4:2:2 is compressed 2.7:1 (negligible with MPEG-4) 4:4:4 is compressed 4.2:1. With twelve channels of audio the data rate to tape is 600mbps in both cases. The SRW5000 is the current studio VTR for the format and it will replay but not record HDCAM. There are daughterboards for this VTR which will provide SD downconversion, HD and SD SDI, 3:2 pulldown in 1080, 720, 625 or 525 and analogue RGB.

 

Phil

[John Sprung]

1) First, it seems that the cameras like the Viper and Dalsa are going in a completely different direction from the Sony/Panasonic world

 

2) Why did Panavision settle on a sort of "middle of the road" with the Genesis? There is one large 35mm size CCD but there is no optical viewfinder and the recording device on the camera cannot record nearly all the resolution the CCD is putting out?

 

3) And finally, what is the mainstream high-end motion picture HD camera system going to most look like in the future

Viper and Dalsa are not at all similar. Viper is a three chip 2/3" camera that has a unique way of dividing pixels up into subpixels vertically (but not horizontally). Dalsa is a big chip camera with a conventional pixel structure. Last time we saw it, it wasn't really ready for production.

 

I'm not sure what's "middle of the road" about the Genesis. It's the first practical big chip camera. Three chips are definitely not better than one if the one chip is big enough. As for the recording format, I'm sure they'll upgrade it as soon as new technology becomes available. That part comes from Sony, Panavision just buys the best they can find. With a good enough display, I'm not a big fan of an optical viewfinder. At the end of a long day, you're in much better shape if you're operating with both eyes open and your head in a natural position. I'd expect the high end cameras of the future to look more like Genesis than anything else that's out there today.

 

 

 

-- J.S.

[Rob van Gelder]

I´m just wondering: from Phil´s post I learn that there are 12 audio channels.

Who needs that in a video-camera? I can understand that the end-product could need that for all the languages etc, but for recording the master material?

Is that bandwith that´s taken by the audio tracks usable for the picture itself?

Or is that hardly any saving/advantage?

 

Rob van Gelder

[Mark Allen]

I´m just wondering: from Phil´s post I learn that there are 12 audio channels.

Who needs that in a video-camera?

12 audio channels has two purposes:

 

1) On movies it is common now to record with DA088's and use all 8 channels for various mics - individual booms, individual lavs, a comp of the whole mix, perhaps even an ambient channel. so, if you're recording data anyway, it makes sense to put all of the tracks in synch immediately.

 

2) For live concerts, these are shot with upwards to 48 or 64 tracks to be mixed later. Some concerts have even mixed to picture depending on who you're looking at. 12 is a lot less, but at least would get you in the range for many concerts.

[Michael Nash]

With a good enough display, I'm not a big fan of an optical viewfinder.  At the end of a long day, you're in much better shape if you're operating with both eyes open and your head in a natural position.

And have you seen the Genesis' viewfinder? ;) The large, color, hi-res Accuscene VF is so good it's almost like looking at ground glass. I think there is still a benefit to seeing what the camera is recording (even for the camera operator, not just the DP), rather than just what passes through the lens. The new VF is really, really good.

[John Sprung]

And have you seen the Genesis' viewfinder?

Yes, that's why I wrote that. It's plenty good enough to operate by, and you can look at a big picture with both eyes. The old parallax BNC finder was like that.

 

 

 

-- J.S.

[Rob van Gelder]

1) On movies it is common now to record with DA088's and use all 8 channels for various mics - individual booms, individual lavs, a comp of the whole mix, perhaps even an ambient channel. so, if you're recording data anyway, it makes sense to put all of the tracks in synch immediately.

 

2) For live concerts, these are shot with upwards to 48 or 64 tracks to be mixed later. Some concerts have even mixed to picture depending on who you're looking at. 12 is a lot less, but at least would get you in the range for many concerts.

I´m sure you mention this out of experience.

I never had any encounter with a soundcrew that works like this.

 

What I mostly see nowadays is still a separate system: sound is recording on their own equipment, Dat, HD, Digital Nagra or whatever. The soundguy does the initial mixing on the set with as many channals as he like or can handle, not sure if this is saved on 2-track (stereo) or more.

 

On live concerts I can imagine this, but again, I´m pretty sure that they will use a separate system to record everything for the mixing later. ( if they are wise)

The sound tracks on the video format are than not much more than guide-tracks so why do you need more than 2-4?

 

Rob van Gelder

[Phil Rhodes]

Hi,

 

I can't beleieve anyone would choose to shoot a concert on something like a Genesis - this is where an ENG-style camera would be fantastic.

 

Phil

[Rob van Gelder]

Exactly Phil, that´s why I question the statements from Mark Douglas.

He writes this as it is common practice, but I still have to see that in my working experience.

 

I did a concert not long ago (Christina Aquiera) in Holland, there were no video camera´s. We were shooting with 9 35mm camera´s, so the "old style" with timecode slates. Sound was seperate, 24+ channels.

 

Rob van Gelder

[Eric Steelberg ASC]

1)  First, it seems that the cameras like the Viper and Dalsa are going in a completely different direction from the Sony/Panasonic world with the optical viewfinder, larger CCD for 35mm depth of field, etc.  Why are Sony and Panasonic not quicker to embrace these film-style characteristics?

 

2)  Why did Panavision settle on a sort of "middle of the road" with the Genesis?  There is one large 35mm size CCD but there is no optical viewfinder and the recording device on the camera cannot record nearly all the resolution the CCD is putting out?

 

 

I have a personal theory that the HD world might split into two eventually.  The ENG/HD world with an electronic viewfinder, 2/3" chips (because a shallow DOF and bigger lenses would be bothersome when shooting a home-improvement show, for example) and a recording devise onboard where compression is acceptable and welcome because it brings with it ease of use and quick work-flow.  And then the high-end cinema HD with optical viewfinder, large CCD, no recording devise on board for the massive amount of data required to store an image that approaches film resolution, and a more complex work flow that will make it harder to deal with the huge high-res data files.  Sound right?

 

I do believe that some day, this high end digital technology will win over the hearts of even the most discerning eye.  That seems to be quite a way in the future.  But what is so funny is watching companies like Dalsa and Viper BENDING OVER BACKWARDS, going to through the greatest lengths to try to be competitive with the current film technology.  And after tons and tons of R & D that Dalsa went through, the best they can try to shoot for at their side by side comparison in Canada is to be "as good as" the newest Vision stocks out there. It is just funny to see people so passionately pursuing a new technology that can only hope to produce results as good as the current technology leader.  Think of how far we are from reaping the benefits of using the Dalsa, fore example, over film.  New untested work flowlow, different post production, unproven and uncertain results.....I guess you have to start somewhere......

It should be noted that Panavision developed the Genesis WITH Sony starting in 2000. In fact, Panavision said the camera is more Sony than Panavision. Panavision designed the body and told Sony what they wanted, Sony developed the technology and made it fit inside.

 

So in a sense, Sony is going in the same direction.

 

To your second point (#2), Genesis can hardly be consiered "middle of the road." Once it's in service, it will be apparent that it is the best all around digital camera, this I'm sure of. The reason they put a such a large sensor in it, whose data is to much to record, is that 1, it provides a superior oversampled image, 2, it enable them to use 35mm lenses, and 3 it future proofs the camera for some time. They spent alot of time and money on this thing and they don't want to have to reinvest time and money every time they want more resolution from a bigger chip. This way they only have to wait for storage technology to improve.

 

I agree with your ideas about ENG style cameras and HD Cinema style camera co-existing in the future.

 

I, too, find it funny to see all of this money has created the pinnacle of video technology which they say delivers unprecedented imagery....as good as the best film stocks. What they don't advertize is how much the cameras cost and how much more the semi-permanent storage costs.

 

I still prefer the robustness of a film camera and integrity of film. HD WILL win...some day.

[Patrick Myles]

Good afternoon.

 

I wanted to join your discussion to address some of the technical points that I have seen coming up in the HD forum over the past couple of months. I?ve read a number of statements about the various imaging technologies that are either incorrect or based on assumptions that don?t necessarily hold water. I think what?s been missing from these discussions is some perspective from folks that are close to the chips and camera technology. Having been with DALSA for a number of years and being close to the roll-out of our Origin camera since before NAB 2003, I?m hoping I can add to the conversation.

 

The first point I want to make has to do with 4K capture. There is a false assumption that other camera companies don?t do 4K capture because of the limitations of the recording technology. That?s just not correct, and has more to do with convenient marketing spin. In reality, the limitation is at the chip and camera level. Building a 4K, professional grade image sensor, CMOS or CCD, with large pixels and 16 bit output that can run at up to 48 frames per second is non-trivial. Designing a camera around that chip with giga-pixel+ capabilities is also a significant challenge.

 

A second related point has to do with David Mullen?s comment about the 12MP CCD in the Genesis camera and how it can probably send out more data than the HDCAM-SR recorder can handle. That may be true to a certain extent, but it?s unlikely that it was ever designed to provide any more than 2K capture at 10-bit log. I think it?s important to cut through the marketing spin and dispel that idea that there is a magic switch that can be turned on to transform the chip into a high dynamic range 4K device once the recording technology catches up ? it simply doesn?t work that way. Part of the problem here is there is little or no technical information available on this Sony chip, so much of the conversation is based on speculation. In case you?re interested in getting into the bits and bytes of our chip and camera, you can find a number of decent technical white papers at http://www.dalsa.com/dc/documents/documents.asp (BTW, that last sentence will be my one and only shameless plug, I promise).

 

Patrick Myles

DALSA

[John Sprung]

The sound tracks on the video format are than not much more than guide-tracks so why do you need more than 2-4?

The video format is the same all the way through production and post. The tracks are there, and the tape transports are made with audio heads, whether you need them for a particular application or not. In the future we may indeed see a compact, rugged MOS HD camera, sort of the 2C of HD. But for the larger sync sound cameras, the decision seems to be driven by the fact that it's just a tiny bit more electronics to give you access to those audio tracks. It's just too easy not to do.

 

 

 

-- J.S.

[David Mullen ASC]

Thanks for the info!

 

I'm curious -- how has the Dalsa changed since I saw at it NAB 2003? Especially in regards to sensitivity and ease of data recording. I recall seeing a data recorder the size of a mini-fridge that could only hold 20 minutes of 4K data. I assume there has been some leaps in that regard.

[John Sprung]

Genesis can hardly be consiered "middle of the road." Once it's in service, it will be apparent that it is the best all around digital camera

While I agree that the Genesis is a magnificent achievement, there is the distinct possibility that it will soon be joined at the top of the hill by the Arri D-20. I'd expect that both will be excellent, and quite different from each other.

 

The big difference is that the Genesis is CCD, and the D-20 is CMOS. This could shake out to be like Kodak vs. Fuji, only more so. On the technical side, CCD and CMOS have substantially different sets of strengths and weaknesses. I'm looking forward to seeing side by side tests so we can judge the aesthetic differences.

 

 

 

-- J.S.

[Patrick Myles]

David, there have been several enhancements to the Origin camera in the past year and a half since NAB '03 (If I'm not mistaken, I think you and I talked briefly at NAB at Geoff Boyle's bash after our launch). The newest generation of our CCD chip that will make its way into the commerically available camera in November will have a "film-speed" of 400 ASA, and enhanced imaging performance.

 

As far as storage, we've gone from the "mini-fridge", as you call it, at NAB '03 (that was a year and a half old technology at the time). That early prototype unit could store 50 minutes. The final recorders, which companies like SGI are developing for us, will be considerably smaller, more ergonomically pleasing and portable, and be able to store up to a couple of hours of data.

[David Mullen ASC]

Will you be working through some rental house to make this camera available, or will you be waiting for a rental house to buy one, or will you be renting it directly? Or is this for sale only?

 

The camera has a 2.00 : 1 image target, correct? And uses standard cine lenses?

[Eric Steelberg ASC]

The first point I want to make has to do with 4K capture. There is a false assumption that other camera companies don?t do 4K capture because of the limitations of the recording technology. That?s just not correct, and has more to do with convenient marketing spin. In reality, the limitation is at the chip and camera level. Building a 4K, professional grade image sensor, CMOS or CCD, with large pixels and 16 bit output that can run at up to 48 frames per second is non-trivial. Designing a camera around that chip with giga-pixel+ capabilities is also a significant challenge.

 

A second related point has to do with David Mullen?s comment about the 12MP CCD in the Genesis camera and how it can probably send out more data than the HDCAM-SR recorder can handle. That may be true to a certain extent, but it?s unlikely that it was ever designed to provide any more than 2K capture at 10-bit log. I think it?s important to cut through the marketing spin and dispel that idea that there is a magic switch that can be turned on to transform the chip into a high dynamic range 4K device once the recording technology catches up ? it simply doesn?t work that way. Part of the problem here is there is little or no technical information available on this Sony chip, so much of the conversation is based on speculation. In case you?re interested in getting into the bits and bytes of our chip and camera, you can find a number of decent technical white papers at http://www.dalsa.com/dc/documents/documents.asp (BTW, that last sentence will be my one and only shameless plug, I promise).

 

Patrick Myles

DALSA

Patrick,

Respectfully, I think it is in your best interest to make it seem like other manufacturers are 'spinning' weaknesses into strengths. I don't believe it is a false assumption to say recording technology is too limited. Of course it is! Your camera is proof. You need to run a fiber cable to a bank of drives (at highest quality output) while hoping it doesn't break WHEN it gets stepped on or WHEN the dolly rolls over it. In fact within the size of the origin there isn't even a tape recording mechanism. Why? Tape can't record it's data and the camera needs that space for processing. That's a recording limitation...the very fact the Origin has no provision for onboard recording. Panavision made onboard, independent, tetherless aquisition it's strength.

 

What I was stating, as others have, including Panavision to me, is that as recording technology increases, so will their ability to fully utilize the chip in the camera without having to change as much as if thyer were stating from scratch. I am the last one to base anything on speculation and only talk about what I've been told. PV has apparently done capturing at 4k in tests and said it's impressive, but 'you can't record to a disk array on steadicam.' No one wants to be a slave to technological limitations on set.

 

All that being said, I AM anxious to see some footage from the Origin. I saw it at Cinegear and was impressed with it's ergonomics and GUI.

[Patrick Myles]

We plan to establish a Hollywood-based partnership to distribute Origin cameras, storage equipment and related accessories on a rental basis, in much the same way as is currently done with film cameras. We recently signed a letter of intent with an established camera rental firm in Los Angeles. We should have a few announcements in the next few weeks.

 

The Origin camera itself has a slightly larger than film sized 8 megapixel (4046 x 2048 pixels) CCD chip that outputs 16-bit linear raw data (12+ stops of exposure latitude) in 4K resolution. It has a PL mount, uses standard cine lenses and has an optical through the lens viewfinder (PS Technik).

[Mitch Gross]

Patrick, nice to see you on the Forum. We met at the Dalsa rollout at NAB as well and I found everyone at Dalsa such as you and John to be very open and forthcoming about the current capablities and limitations of the Origin's design.

 

Has the camera's physical form changed much? I saw John at a trade show in NYC earlier this year and he said that Dalsa was hoping to get more than a 14-stop range by the camera was released next year. When is the Origin due?

 

Always interested.

[Phil Rhodes]

Hi,

 

Is this the guy who didn't know what zebra stripes were?

 

Phil

[Patrick Myles]

Eric, I hope that you will soon have a chance to see the footage from our camera and to get your hands on it when its available starting in November.

 

BTW, I agree that tetherless acquisition is indeed a strength of the Genesis, as is the move to a film-sized chip, allowing you to use standard lenses. These two pieces are definitely a step forward for 2k capture. Obviously 2K is not the direction DALSA went in. Our strong belief is that 4K capture (digital and/or scanned film) is where things are going. It?s quite amazing to see how things have progressed, even in the past few months wrt 4K. Just last week at IBC, SGI and Discreet were demonstrating viable workflows for 4K color grading, using our footage from ?Le Gant/The Glove?. I remember at NAB 2003 there really wasn?t anything that could even reliably capture 2K.

 

With regards to my original point about "spin", image sensor and camera data specs often need to be taken with a grain of salt, particuarly when documentation is short on specifics, like in this case. I'm not questioning whether they have x number of pixel elements, but it's unlikely that the device as it exists now was designed for 4K full resolution output at 24+ fps, and that they are just waiting for storage to catch up. It doesn't fit into the design spec, which is a film-sized chip, 2K camera of a moderate size with on-board storage. A more likely scenario is that the chip was designed specifically to run at 2 megapixels and that the large number of small pixel elements were designed to be ganged together for color fidelity reasons (that's where the striped RGB design comes in) and of course to end up with good 10-bit 2K, a standard format. Like the Viper before it, which touted 9 million pixels per chip (it was actually 2 million as John Sprung thankfully pointed out), big numbers always sound better. Maybe we should get rid of all those marketing and PR people (oops, I guess that would include me!).

 

WRT fibre optic cabling that connects our camera to a storage device, FO cable is rugged tactical cable designed for military use. We tried to break it to confirm and we were rolling over with a loaded fork lift, dolly, driving the car over it and all kinds of "nice" things like this. The ultimate test came while we were shooting in Montreal, the cable got caught on the dolly track on a long track shot and the moving dolly (with two people on it) was nicely stopped. No damage was done to the camera, cable (nor the track....).

[Eric Steelberg ASC]

Eric, I hope that you will soon have a chance to see the footage from our camera and to get your hands on it when its available starting in November.

 

BTW, I agree that tetherless acquisition is indeed a strength of the Genesis, as is the move to a film-sized chip, allowing you to use standard lenses. These two pieces are definitely a step forward for 2k capture. Obviously 2K is not the direction DALSA went in. Our strong belief is that 4K capture (digital and/or scanned film) is where things are going. It?s quite amazing to see how things have progressed, even in the past few months wrt 4K. Just last week at IBC, SGI and Discreet were demonstrating viable workflows for 4K color grading, using our footage from ?Le Gant/The Glove?. I remember at NAB 2003 there really wasn?t anything that could even reliably capture 2K.

 

With regards to my original point about "spin", image sensor and camera data specs often need to be taken with a grain of salt, particuarly when documentation is short on specifics, like in this case. I'm not questioning whether they have x number of pixel elements, but it's unlikely that the device as it exists now was designed for 4K full resolution output at 24+ fps, and that they are just waiting for storage to catch up. It doesn't fit into the design spec, which is a film-sized chip, 2K camera of a moderate size with on-board storage. A more likely scenario is that the chip was designed specifically to run at 2 megapixels and that the large number of small pixel elements were designed to be ganged together for color fidelity reasons (that's where the striped RGB design comes in) and of course to end up with good 10-bit 2K, a standard format. Like the Viper before it, which touted 9 million pixels per chip (it was actually 2 million as John Sprung thankfully pointed out), big numbers always sound better. Maybe we should get rid of all those marketing and PR people (oops, I guess that would include me!).

 

WRT fibre optic cabling that connects our camera to a storage device, FO cable is rugged tactical cable designed for military use. We tried to break it to confirm and we were rolling over with a loaded fork lift, dolly, driving the car over it and all kinds of "nice" things like this. The ultimate test came while we were shooting in Montreal, the cable got caught on the dolly track on a long track shot and the moving dolly (with two people on it) was nicely stopped. No damage was done to the camera, cable (nor the track....).

Interesting about the cable. Hadn't heard that development yet.

 

I agree with everything you have said and look forward to seeing how each format is adopted.

[Balazs Rozsa]

In reality, the limitation is at the chip and camera level. Building a 4K, professional grade image sensor, CMOS or CCD, with large pixels and 16 bit output that can run at up to 48 frames per second is non-trivial.

Hello Patrick,

 

I'm glad to see a person here from DALSA. Can I ask why it is that much more problematic to design a 4K sensor than a smaller 2K one. At first look I would think that if you put together 4 pieces of 2K sensors, you will have a 4K sensor. You need to increase the number of signal outputs four times as well to make the read-out speed large enough because of the increased number of pixels. Is the production yield problematic? I'm very curious. Thanks,

 

Balazs